Decision Support Tool for Self-Management of Insulin Dosing in Type 1 Diabetes with Multiple Daily Injections: A Proof-of-Concept Study

Abstract

Objective:

To assess feasibility and safety of a decision support system (AI-DSS) that provides algorithm-generated insulin dosing recommendations directly to individuals with type 1 diabetes (T1D) managed with multiple daily injections (MDI).

Methods:

This single-arm, prospective proof-of-concept study included individuals with T1D managed with MDI and continuous glucose monitoring (CGM). Participants underwent a 4-week run-in period followed by a 12-week intervention phase, during which every two weeks algorithm-generated insulin titration recommendations were provided via a mobile application. CGM metrics were compared between the last 2 weeks of the run-in (baseline) and the last 2 weeks of the intervention periods. Primary safety outcomes included percent time <54 mg/dL and >250 mg/dL. Secondary outcomes included changes in HbA1c and time in range (TIR, 70–180 mg/dL).

Results:

The study cohort included 16 young adults (mean age 25.1 ± 4.1 years; 56% female, mean HbA1c 7.6% ± 0.8%) who completed the study. Median HbA1c significantly decreased from 7.5% (IQR: 7.1, 8) to 7.1% (IQR: 6.5, 7.3), from start to end of study (P = 0.013). TIR significantly improved by 3.5% ± 7.3% (P = 0.039). Time <54 mg/dL remained unchanged (0.9% ± 0.86% vs. 1.12% ± 1.11%; P = 0.191), with a trend toward reduced time >250 mg/dL (14.3% ± 10.71% vs. 12.32% ± 10.91%; P = 0.055). No severe adverse events were reported.

Conclusion:

Decision support tool for self-managed insulin dosing in individuals with T1D using MDI was feasible, safe, and improved glycemic control, supporting further evaluation in large-scale randomized trials.